[Development] [FYI] new git-gpush features, a.k.a. the smart way of pushing to gerrit

Christian Kamm kamm at incasoftware.de
Fri Oct 24 14:01:29 CEST 2014

Jędrzej Nowacki wrote:
> Sadly It is because of two things:
> 1. It crates new patchset, which sends notification and reset score
> 2. It breaks web diff.
> Both should be fixable on the Gerrit level

I agree that rebases are currently very frustrating for reviewers and
that you've spelled out the root problems. I suggest to

1. Fix gerrit to generate useful interdiffs between patchsets that hide
unrelated changes between the base revisions. As Olivier points out I've
done that in https://github.com/ckamm/gerrit/tree/interdiff-stable-2.8.

2. Use a smarter version of the trivial rebase script to propagate
reviews to new patchsets. Ideally that'd use the better diffs gerrit
emits for its decision. I haven't done that.

My implementation of 1 is a hack, definitely not perfect and doesn't
work at all with gerrit's new change screen yet. But it's worked well
for me on top of gerrit 2.8 for more than a year. If Digia/QtCompany is
interested, I may be able to help out with integration or any further
fixes that become necessary in that area.

Christian Kamm

More information about the Development mailing list